[Bug fortran/47065] New: Replace trim(a) by a(1:len_trim(a))

[Bug fortran/47065] New: Replace trim(a) by a(1:len_trim(a))

[tkoenig at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47065

           Summary: Replace trim(a) by a(1:len_trim(a))
           Product: gcc
           Version: 4.6.0
            Status: UNCONFIRMED
          Keywords: missed-optimization
          Severity: enhancement
          Priority: P3
         Component: fortran
        AssignedTo: unassigned@gcc.gnu.org
        ReportedBy: tkoenig@gcc.gnu.org


In many cases, it is possible to replace trim(a) by a(1:len_trim(a)).

This is a win because it saves trim from creating a temporary.

[Bug fortran/47065] Replace trim(a) by a(1:len_trim(a))

[tkoenig at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47065

Thomas Koenig <tkoenig at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |ASSIGNED
   Last reconfirmed|                            |2010.12.26 21:31:55
         AssignedTo|unassigned at gcc dot       |tkoenig at gcc dot gnu.org
                   |gnu.org                     |
     Ever Confirmed|0                           |1

[Bug fortran/47065] Replace trim(a) by a(1:len_trim(a))

[tkoenig at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47065

--- Comment #1 from Thomas Koenig <tkoenig at gcc dot gnu.org> 2010-12-26 21:34:14 UTC ---
Created attachment 22852
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=22852
Partial patch

This appears to work for the cases where there is no substring reference.

[Bug fortran/47065] Replace trim(a) by a(1:len_trim(a))

[tkoenig at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47065

--- Comment #2 from Thomas Koenig <tkoenig at gcc dot gnu.org> 2010-12-31 11:32:20 UTC ---
Author: tkoenig
Date: Fri Dec 31 11:32:16 2010
New Revision: 168367

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=168367
Log:
2010-12-31  Thomas Koenig  <tkoenig@gcc.gnu.org>

    PR fortran/47065
    * frontend-passes.c (count_arglist):  Static variable to
    count the nesting of argument lists.
    (optimize_code):  Set count_arglist to 1 if within a call
    statement, to 0 otherwise.
    (optimize_trim):  New function.
    (optimize_expr):  Adjust count_arglist.  Call optimize_trim.

2010-12-31  Thomas Koenig  <tkoenig@gcc.gnu.org>

    PR fortran/47065
    * gfortran.dg/trim_optimize_3.f90:  New test.
    * gfortran.dg/trim_optimize_4.f90:  New test.


Added:
    trunk/gcc/testsuite/gfortran.dg/trim_optimize_3.f90
    trunk/gcc/testsuite/gfortran.dg/trim_optimize_4.f90
Modified:
    trunk/gcc/fortran/ChangeLog
    trunk/gcc/fortran/frontend-passes.c
    trunk/gcc/testsuite/ChangeLog

[Bug fortran/47065] Replace trim(a) by a(1:len_trim(a))

[tkoenig at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47065

--- Comment #3 from Thomas Koenig <tkoenig at gcc dot gnu.org> 2010-12-31 11:38:03 UTC ---
The most common cases are handled now.

Still to do:

- Handle substring references, for example trim(a(3:5)). Here, we have
  to watch out for trim(a(function(x):5)), not to evaluate the function
  twice.

- Right now, the optimization is not applied in argument
  lists.  This can be relaxed for PURE functions and subroutines,
  or if dependency analysis shows that it cannot hurt. The latter point
  is probably not worth it.

[Bug fortran/47065] Replace trim(a) by a(1:len_trim(a))

[burnus at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47065

Tobias Burnus <burnus at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |burnus at gcc dot gnu.org

--- Comment #4 from Tobias Burnus <burnus at gcc dot gnu.org> 2011-01-18 21:03:06 UTC ---
(In reply to comment #3)
>   This can be relaxed for PURE functions and subroutines,
And for implicitly pure procedures (attr.implicit_pure).

[Bug fortran/47065] Replace trim(a) by a(1:len_trim(a))

[tkoenig at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47065

--- Comment #5 from Thomas Koenig <tkoenig at gcc dot gnu.org> 2011-01-18 23:21:06 UTC ---
Created attachment 23024
  --> http://gcc.gnu.org/bugzilla/attachment.cgi?id=23024
patch handling references

This patch would work, but for PR 47348.

[Bug fortran/47065] Replace trim(a) by a(1:len_trim(a))

[tkoenig at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47065

--- Comment #6 from Thomas Koenig <tkoenig at gcc dot gnu.org> 2011-03-27 17:40:29 UTC ---
Author: tkoenig
Date: Sun Mar 27 17:40:26 2011
New Revision: 171575

URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=171575
Log:
2011-03-27  Thomas Koenig  <tkoenig@gcc.gnu.org>

    PR fortran/47065
    * frontend-passes (optimize_trim): Also follow references, except
    when they are substring references or array references.

2011-03-27  Thomas Koenig  <tkoenig@gcc.gnu.org>

    PR fortran/47065
    * gfortran.dg/trim_optimize_5.f90:  New test.
    * gfortran.dg/trim_optimize_6.f90:  New test.


Added:
    trunk/gcc/testsuite/gfortran.dg/trim_optimize_5.f90
    trunk/gcc/testsuite/gfortran.dg/trim_optimize_6.f90
Modified:
    trunk/gcc/fortran/ChangeLog
    trunk/gcc/fortran/frontend-passes.c
    trunk/gcc/testsuite/ChangeLog

[Bug fortran/47065] Replace trim(a) by a(1:len_trim(a))

[fxcoudert at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47065

--- Comment #7 from Francois-Xavier Coudert <fxcoudert at gcc dot gnu.org> 2011-11-09 10:09:15 UTC ---
Thomas, I think this is mostly fixed, isn't it? If so, we could close it.

[Bug fortran/47065] Replace trim(a) by a(1:len_trim(a))

[tkoenig at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47065

Thomas Koenig <tkoenig at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED

--- Comment #8 from Thomas Koenig <tkoenig at gcc dot gnu.org> 2013-03-17 15:26:18 UTC ---
I think this is fixed as far as it is useful.

Marking as closed.